* Fix error-prone SQL queries in Account search
While this code seems to not present an actual vulnerability, one could
easily be introduced by mistake due to how the query is built.
This PR parameterises the `to_tsquery` input to make the query more robust.
* Harden code for Status#tagged_with_all and Status#tagged_with_none
Those two scopes aren't used in a way that could be vulnerable to an SQL
injection, but keeping them unchanged might be a hazard.
* Remove unneeded spaces surrounding tsquery term
* Please CodeClimate
* Move advanced_search_for SQL template to its own function
This avoids one level of indentation while making clearer that the SQL template
isn't build from all the dynamic parameters of advanced_search_for.
* Add tests covering tagged_with, tagged_with_all and tagged_with_none
* Rewrite tagged_with_none to avoid multiple joins and make it more robust
* Remove obsolete brakeman warnings
* Revert "Remove unneeded spaces surrounding tsquery term"
The two queries are not strictly equivalent.
This reverts commit 86f16c537e06c6ba4a8b250f25dcce9f049023ff.
* Fix followers synchronization mechanism not working when URI has empty path
To my knowledge, there is no current implementation on the fediverse
that can use bare domains (e.g., actor is at https://example.org instead of
something like https://example.org/actor) that also plans to support the
followers synchronization mechanism. However, Mastodon's current implementation
would exclude such accounts from followers list.
Also adds tests and rename them to reflect the proper method names.
* Move url prefix regexp to its own constant
* Fix error when suspending user with an already-existing canonical email block
Fixes#17033
While attempting to create a `CanonicalEmailBlock` with an existing hash would
raise an `ActiveRecord::RecordNotUnique` error, this being done within a
transaction would cancel the whole transaction. For this reason, checking for
uniqueness in Rails would query the database within the transaction and avoid
invalidating the whole transaction for this reason.
A race condition is still possible, where multiple accounts sharing a canonical
email would be blocked in concurrent transactions, in which only one would
succeed, but that is way less likely to happen that the current issue, and can
always be retried after the first failure, unlike the current situation.
* Add tests
For some reason, some misconfigured servers return an empty document when
queried over webfinger. Since an empty document does not lead to a parse
error, the error is not caught properly and triggers uncaught exceptions
later on.
This PR fixes that by immediately erroring out with `Webfinger::Error` on
getting an empty response.
Fixes#16515
Not using a router object somehow made `this.history` lag behind the real
browser history whenever pushing a new history item in `replyCompose`.
Not using the context-provided router in this case was an oversight made
when porting glitch-soc changes in #16499.
Up until now, we have used Devise's Rememberable mechanism to re-log users
after the end of their browser sessions. This mechanism relies on a signed
cookie containing a token. That token was stored on the user's record,
meaning it was shared across all logged in browsers, meaning truly revoking
a browser's ability to auto-log-in involves revoking the token itself, and
revoking access from *all* logged-in browsers.
We had a session mechanism that dynamically checks whether a user's session
has been disabled, and would log out the user if so. However, this would only
clear a session being actively used, and a new one could be respawned with
the `remember_user_token` cookie.
In practice, this caused two issues:
- sessions could be revived after being closed from /auth/edit (security issue)
- auto-log-in would be disabled for *all* browsers after logging out from one
of them
This PR removes the `remember_token` mechanism and treats the `_session_id`
cookie/token as a browser-specific `remember_token`, fixing both issues.
* Refactor shouldUpdateScroll passing
So far, shouldUpdateScroll has been manually passed down from the very top of
the React component hierarchy even though it is a static function common to
all ScrollContainer instances, so replaced that with a custom class extending
ScrollContainer.
* Generalize “press back to close modal” to any modal and to public pages
* Fix boost confirmation modal closing media modal
* Change references to tootsuite/mastodon to mastodon/mastodon
* Remove obsolete test fixture
* Replace occurrences of tootsuite/mastodon with mastodon/mastodon in CHANGELOG
And a few other places
* Fix anonymous access to outbox not being cached by the reverse proxy
Up until now, anonymous access to outbox was marked as public, but with a
0 duration for caching, which means remote proxies would only serve from cache
when the server was completely overwhelmed.
Changed that cache duration to one minute, so that repeated anonymous access
to one account's outbox can be appropriately cached.
Also added `Signature` to the `Vary` header in case a page is requested, so
that authenticated fetches are never served from cache (which only contains
public toots).
* Remove Vary: Accept header from webfinger controller
Indeed, we have stopped returning xrd, and only ever return jrd, so the
Accept request header does not matter anymore.
* Cache negative webfinger hits for 3 minutes
* Add tests
* Fix serialization of followers/following counts when user hides their network
Fixes#16382
Signed-off-by: Claire <claire.github-309c@sitedethib.com>
The auto-linking code basically rewrote the whole string escaping non-ascii
characters in an inefficient way, and building a full character offset map
between the unescaped and escaped texts before sending the contents to
TwitterText's extractor.
Instead of doing that, this commit changes the TwitterText regexps to include
valid IRI characters in addition to valid URI characters.
Fixes#16509
Microsoft Edge with translation enabled rewrites the DOM in ways that confuse
react and prevent it from working properly. Wrapping the offending parts in
a span avoids this issue.
* Add tests
* Fix some link previews being incorrectly generated from different prior links
PR #12403 added a cache to avoid redundant queries when the OEmbed endpoint can
be guessed from the URL. This caching mechanism is not perfectly correct as
there is no guarantee that all pages from a given domain share the same
OEmbed provider endpoint.
This PR prevents the FetchOEmbedService from caching OEmbed endpoint that
cannot be generalized by replacing a fully-qualified URL from the endpoint's
parameters, greatly reducing the number of incorrect cached generalizations.
* Add tests
* Add security-related tests
My first (unpublished) attempt at fixing the issues introduced (extremely
hard-to-exploit) security vulnerabilities, addressing them in a test.
* Fix authentication failures after going halfway through a sign-in attempt
* Refactor `authenticate_with_sign_in_token` and `authenticate_with_two_factor` to make the two authentication steps more obvious
